Brain tumors are one of the more common cancers in individuals under the age of 35 and over the age of 60. For the vast majority of patients survival is still very poor. One of the major problems with stratifying patients who will respond to therapy or who have a better prognosis is the heterogeneity that often surrounds the histopathological classification of these tumors. Even highly trained neuropathologists often disagree on the diagnosis, which means that entering patients on specific trials can be hit-and-miss. Furthermore, within tumor types some patients do well and others do not. This observation suggests that there are underlying genetic events that determine the clinical outcome which cannot be classified on histopathology alone. What is clearly needed is a more objective way of stratifying brain tumor patients. Cytogenetic analysis of brain tumors has provided some clues as to the underlying genetic heterogeneity in brain tumors but the resolution of these approaches is inadequate to reveal the majority of genetic events that are present. It is likely, based on our limited knowledge to date, that brain tumors acquire small deletions and amplifications of critical genes during their development. Our hypothesis is that these changes could be the basis on which to subclassify brain tumors. The recent development of BAC arrays at RPCI now allows a high-resolution comparative genome hybridization (CGHa) analysis of tumor DNA which has the ability to detect deletions and amplifications currently down to 750Kb in a single hybridization experiment. We propose to use this novel technology to screen over 150 brain tumors for genetic changes and then relate the genetic changes to clinical parameters. Ultimately we will be able to develop a custom array which will allow an unbiased analysis of brain tumors in an attempt to predict clinical phenotypes such as response to therapy and prognosis.